1 

FINAL

Dugong Aerial Survey Report

May 25-29, 2008

Bazaruto Archipelago National Park Inhambane Province, Mozambique

World Wide Fund for Nature

December 2008

Prepared by:

Jane A Provancha and Eric D. Stolen

Dynamac Corporation Kennedy Space Center, FL, U.S.A

2 

Introduction:

The World Wide Fund for Nature is actively involved with stakeholders in the

conservation planning and implementation for the Bazaruto Archipelago National Marine

Park in the Inhambane province of Mozambique, east Africa. Several species of

regional and global importance are known to inhabit this area including dugongs

(Dugong dugon). These marine mammals are members of the order Sirenia (manatees

and dugongs) all species of which are listed by the IUCN as either threatened or

endangered. Dugongs are represented by only one species and while they are

abundant along the coasts of Indonesia and Australia they are in apparent decline along

the east African coast. Various reports have lead to the suggestion that Western Indian

Ocean dugongs may now remain in only small numbers in areas of Kenya, Tanzania,

Mozambique, Madagascar, Seychelles, and the Comoros archipelago. Dugongs found

in the Bazaruto Archipelago, Mozambique are considered to be the only viable dugong

population within the entire Western Indian Ocean (Marsh et al. 2006, Dutton 1994).

Cockcroft et al. (2008) reviewed recent records and indicated that few dugongs occur

elsewhere on the Mozambique coast. An aerial census in May 2001 of the Bazaruto

National Park and the eastern islands conducted by (Mackie/WWF 2001) found

dugongs distributed throughout the northern, central and south areas of the Archipelago

between Bazaruto Island and the mainland.

Reports of dugong abundance vary widely for this area. Dutton (1998)

conducted aerial surveys that yielded only 21 dugongs and he suggested dugongs were

rapidly declining based in these low counts. Guissamulo & Cockcroft (1997) estimated

a total local population of 130 dugongs in the Bay based on strip transect aerial surveys.

WWF (2004) reviewed various aerial counts with estimates of 20 to 130 individuals

between 1990 and 2002, and also suggested that the population is in decline (Mackie,

1999). Unfortunately details of some of the surveys are not clear in terms of methods,

areal extent and conditions, making comparison of animal numbers or trends very

difficult and in some cases impossible. In May 2008, WWF initiated what would be an

annual systematic dugong monitoring program for the BANP to assess general

distribution and use within and nearby the park. This effort also involved training Park

3 

and WWF staff to begin the regular aerial monitoring program for Mozambique’s

dugongs. This report describes the methods and outcome of the 2008 WWF surveys.

Figure 1. Bazaruto Archipelago National Park vicinity map with the park boundaries highlighted with yellow squares. (courtesy BANP Integrated Management Plan, 2006). Methods

Systematic aerial surveys were performed utilizing standard techniques including

a trained survey team familiar with dugongs and other species common to the Bazaruto

region. The principal investigator (Provancha) has over 25 years experience in aerial

surveys for manatees, sea turtles and dolphins in estuarine and marine systems. Initial

reconnaissance surveys and team training were performed from 23 – 25 May which

allowed for familiarization of conditions in the area, determination of realistic survey

4 

flight path and time periods, preparation of observers, and recorder for efficient data

communication. Timing considerations included a variety of environmental variables,

including tides, sunlight penetration of the waters, cloud cover, wind, etc. Winds greater

than 15-18 kts, creating high sea-states were considered unacceptable conditions for

surveys.

A Cessna 206 aircraft provided the aerial platform for the dugong surveys and

included a pilot experienced in many hours of animal surveys the southeastern region of

Africa. The aircraft seated seven and included communication system (headsets) to

enhance communication efficiencies between the pilot and all team members. It was

also outfitted with standard GPS systems and a PDA loaded with ESRI Arcpad to log

the flightpath associated with each survey.

Study area and design

Based on reports of numerous dugong sightings north of the BANP, the surveys

included the Indian Ocean waters east of Bartolomeu Dias ( S 21 º deg 09``) and south

through the BANP to Cabo S. Sebastião ( S22º 13``). The southern section originally

included the shallow bay south of Vilanculos to Cabo S. Sebastião. Cockroft et al.

(2008) performed flights over this bay but did not observe any dugongs there, however,

Guissamulo (April 2008, pers. comm.) suggested many of those flights were aborted

due to aircraft control issues at the nearby Vilankulos airport. The quiet waters of this

bay were of interest to the management of the population of dugongs as a whole

therefore, transects for this subsection were planned and oriented to accommodate

airport traffic concerns by reducing time passing over this airspace.

After reviewing the study area, distance between transects was set at 4 km in

order to yield reasonable observation coverage from an altitude of 500 feet (152m).

The transect lines generally oriented east and west and could be considered one

continuous flight track as we also searched for dugongs along the north south legs

connecting the transects, Figure 2. This design involved 26 east-west transects and 25

north-south legs. The total original flight path was over 640km with transect spacing

intended to allow for survey completion within a four and a half hour period. The initial

5 

design was considered tentative depending on actual effort (air time, observer fatigue,

etc.) experienced in the preliminary reconnaissance and training flights.

Bazaruto Bay and vicinity waters were described by Cockcroft and Guissamulo

(2008) as three distinct basins including one to the north of Santa Carolina Island with a

maximum depth 33 m. This very broad and deepest basin is the main connection of the

bay to the open sea. In the middle section, west of the southern end of Bazaruto Island

and northern end of Benguerra Island, there is a variety of relief with shallow shoals and

deeper areas with maximum depths of about 24 m. The southern bay is comprised of

vast areas of tidal flat areas often totally exposed during spring low tides. These flats

do not support the large expanses of seagrasses or algae found in slightly deeper areas

that are not so consistently exposed.

6 

Figure 2. Study area with original flight path over Bazaruto Bay for May 2008 WWF flights

7 

The pilot was familiar with flight lines based on preflight briefings and utilized the

PDA with GPS and ArcPAD display for navigation along the predetermined flightpath.

Survey data were recorded using ArcPad also running on an 8 x 11 inch Scribbler

(Compaq) computer tablet with a Bluetooth GPS and a customized survey program

(NASA/KSC manatee surveys). The BANP area GIS shapefiles were compiled and

loaded in ArcPad including bathymetric and seagrass layers for viewing during survey

operations. This “manatee entry form” eased the real time recording of multiple

sighting records. Start time, aircraft identification number, fuel level, and tach-hours

were recorded at the outset of each flight. Begin and end time of the actual survey

track was also recorded. Observers were positioned in the two back seats and the

recorder sat in front next to the pilot. For fuel efficiency the aircraft carried only 4 people

during the surveys.

The observers were the official data collectors due to their consistent, diligence in

searching the waters. Rare sighting events announced by the pilot or recorder were

verified by an observer and only included in the output if the observer was confident he

or she would have seen the animal without the initial notification by the non-observer.

The level of effort by the pilot and recorder are typically highly variable and may bias the

count/ index with numbers that would not normally be attained upon repeated surveys.

However, for this sparsely distributed, important species, we captured all verified

observations, as records of nonconforming data points were considered important.

Observers searched out the window between the bottom of the plane out to the

edge of their ability to perceive dugongs or signatures of dugongs (i.e. fluke splash,

feeding plumes, etc.). They also reported the sighting angle of the initial observation

using an Asuunto clinometer. These records allowed for sample distance calculations

of each sighting and the estimated swath width of the observation strip on the sea

surface. Water and sighting conditions were recorded at the beginning of the flight and

updated as needed: (turbidity, sea-state, etc.). Observers informed the recorder of

dugong sightings and included number of adults, calves, behavior (traveling, feeding,

surface rest, etc.).

8 

The project effort focused on enumeration of dugongs but other information

useful to Park management and dugong conservation were also collected as

appropriate. These included numbers and locations of sea turtles, dolphins, and boats.

The turtles and dolphins were not recorded to the level of species although some were

identifiable. Number of individuals per event and sighting angle were collected for

turtles and dolphins. Additionally calves accompanying dolphins were recorded. Boat

categories included sailboats (touristic), motor power boat (touristic), local

fishing/seining boats, and fish traps (gamboas) were also enumerated.

In the event the computer recording system failed (i.e. low battery, etc.), back up

paper maps were onboard for scribing sightings. In this case, the recorder used the

maps and coordinates from the GPS.

Analyses

Automated data were downloaded and quality screened as well as compared to

paper maps as necessary. Outputs of all sighting distributions for each flight were

placed on maps and tables. Data were also analyzed using program DISTANCE

version 5.0, release 2 (Thomas et al. 2006). The general logic of the analysis was to

first estimate the density of dugong clusters by modeling the detection rates for

sightings as a function of distance. Abundance was calculated by multiplying the

estimated cluster density by the expected cluster size and the area of the study site.

Observers suggested that the assumed detectability of surfacing turtles, dugongs, and

dolphins were similar given the water clarity conditions within the study area and the

surfacing requirements of these species. The analysis assumes the observers were

able to detect all large surface-breathing marine vertebrates on the transect center line,

thus all estimation followed the assumption that g(0)=1 was met.  Data were pooled over

the 3 official survey days (27-29 May). Following data screening, we right truncated

detection distances at 5% of the data (400m). We next evaluated several different

models of the detection function using program DISTANCE version 5.0, following

methods described by Buckland et al. (2001). We selected an individual model or set of

best models based on the relative Akaike’s Information Criterion (AIC) value (Burnham

and Anderson 1998) and analysis of the model fit.  The expected cluster size was

estimated two ways: 1) using a size-biased regression method and 2) as the mean

9 

cluster size for all groups seen within 400m of the adjusted centerline. The regression

method is recommended to correct for size-bias in estimation of cluster size, because

larger clusters may be more easily detected at greater distances than are smaller

clusters (Buckland et al. 2001). Expected cluster size and abundance was estimated as

the total number of animals which included adults and calves.

Because the number of observations of dugongs were limited during this survey,

we combined dugong sightings with sightings of sea turtles and dolphins, and used

these data when fitting detection functions. Dugongs, adult sea turtles, and dolphins

are all of similar size, and during preliminary surveys observers noted that all three

appeared to be equally visible in these waters. Differences in the detectability between

dugongs, dolphins and sea turtles were investigated by incorporating the animal type as

a covariate into models of detection function (Marques and Buckland 2004). We also

investigated the potential effects of survey day, observer and cluster size using

covariates in models of the detection function. Following selection of the best model for

the detection function using the combined animal data, we calculated the density of

dugongs only, using the formula LsEfnD

⋅⋅⋅

=2

)(ˆ)0(ˆˆ, where n = the number of dugong

clusters observed within the truncation distance, )0(f̂ = the value of the probability

density function of perpendicular distances evaluated at zero distance (obtained from

the detection function fit to data for all animals), )(ˆ sE = the expected cluster size for

dugongs, estimated from the observed clusters within the truncation distance, and L =

the total combined length of all transects. Variance in D̂ was calculated using the delta

method, as recommended by Buckland et al. (p. 52, 2001). Abundance was calculated

as AD ⋅ˆ , where A = area of study site (174,900 ha).

Due to the aircraft design, there was an area beneath the plane that was not

visible to the observers. Based on measurements on the ground and examination of

sighting data, it was determined that objects with sighting angles greater than 42

degrees from the horizon were not consistently visible during flight. At the designated

survey altitude of 152m, this equated to a distance of 169 m from the centerline on each

10 

side. During analysis of the data, we adjusted observed distance by subtracting 169 m

from all calculated distance, in effect moving the centerline over to the closest position

reliably observed during flight. During surveys, observers recorded some observations

with sighting distances closer to the centerline (e.g. sighting angles greater than 42

degrees), but we later discarded all observations with sighting angles greater than 42

degrees. This method is recommended to avoid introducing more uncertainty (Buckland

et al. 2001). RESULTS

Ground based training sessions occurred on May 23 and 24 incorporating

clinometer readings, observation callouts, and computer program operation. The

reconnaissance survey with experienced personnel occurred on 25 May and included

full data collection along the entire study area. The remaining survey crew (new

trainees) flew on surveys and became acclimated by performing observations and data

entry over the survey area on 26 May. Training flights totaled 7.56 hours of search

time. The subsequent surveys, utilizing the experienced and trained observers,

occurred on 27-29 May and these official surveys resulted in a total search effort of

10.57 hours.

The 25 May reconnaissance flight included the collection of observations along

the seaward edge of the islands for whale sharks and the bay of Cabo S. Sebastião for

dugongs on a low tide. The flight track along outer- ocean was dropped from

subsequent surveys due to effort (time) required. Cabo S. Sebastião bay was surveyed

again on a high tide on 26 May with no dugongs sighted on either flight. Therefore the

area was considered too labor and fuel intensive and so was also cut from subsequent

flight paths. The flight line on Figure 2 that runs parallel and east of the islands and the

3 lines near Cabo Sebastião were eliminated from the study for the official surveys.

This resulted in the survey totaling 23 east-west transects with 22 north south legs and

yielding 521 km flightpath. Sighting events from the aerial surveys are displayed on

figures 3 through 7, and in the Appendix.

Dugong sightings were highly variable with a total of 46 dugong groups

comprising 218 dugongs during the five day period. The dugong counts for the three

11 

“official” surveys were: 9, 22, and 135 for each respective day from May 27 through May

29 (30 sighting events). Table 1a shows the counts and densities (animals/km2) of

dugongs associated with each survey day. Note that on 25-26 May, full surveys were

performed, but with crews “in-training” and so are highlighted in the table but were

eliminated from the population and distance analyses.

Table 1a. Total effort and counts of dugongs each survey day, including training days which are highlighted in grey. DATE Hours Length

(Linearkm)

Count Dugong /Linaer km

Observed Density- (416km2)

Area

May 25* 4.1 700 4(1) 0.006 NA BD-Vil, plus SS & Sea May 26* 3.5 626 48(1) 0.08 NA BD-Vil, plus SS May 27 3.2 521 9(1) 0.02 0.02 BD-Vil May 28 4.1 521 22 (2) 0.04 0.05 BD-Vil May 29 3.4 521 135 (6) 0.26 0.32 BD-Vil

The distribution of dugongs for the five survey days is shown in figure 3, and

while statistical comparisons of data for 25 and 26 May were not made, the distribution

of dugongs sighted those days is considered important. About 38% of the dugong

sightings were outside of the park boundaries. The largest single dugong group (over

70) was sighted in the northern end near the Govuro and Inhassoro boundaries (see the

map for May 29 on figure 7). This group was sighted within 500 meters of a fishing boat

that was tending what appeared to be a shark net. These nets are large mesh

tanglenets capable of drowning dugongs and other air breathers.

12 

Figure 3. Map of dugong distribution within and the vicinity of BANP for all five days combined (May 25- May 29, 2008).

13 

Figure 4. Daily distributions of dugongs within the 2008 dugong aerial survey area in BANP. Numbers indicate counts of adults

ollowed by counts of calves.

14 

Sighting distributions of fishing boats for all five flights combined are shown in

Figure 5. Fishing boats were the dominant vessels observed and were distributed

throughout the region with a minor break in densities near the northernmost end of the

Inhassoro district. The distribution of the other boats for the same period is shown in

Figure 6.

Figure 5. Map of the combined distribution of fishing boats during the five flights from May 25-29, 2008.

15 

Figure 6. Map of the combined distribution of other boats (power or ski boats and touristic sail boats) during the five flights from May 25-29, 2008.

A total of 154 turtles were sighted during the surveys (Table 1b) with sightings

distributed across the entire study area and as shown in Figure 7, most animals were

found in the north half between Bazaruto Island up to Bartolomeu Dias. The individual

turtle sighting events ranged from 36 to 54 for each survey day and are displayed in

16 

Figure A-1. The majority of the turtles appeared to be Green turtles (Chelonia mydas)

and at least three Leatherbacks (Dermochelys coriacea) were also recognized.

Figure 7. Map of sea turtle distribution within and near the BANP for all days combined from 25-29 May, 2008.

17 

Dolphins were observed each day but in lower numbers than expected with totals ranging from 5 to 13. Table 1b indicates the counts for each flight for dolphins. Figure 8. Map of dolphin distribution from aerial surveys conducted for five days from 25-29 May, 2008.

18 

Table 1b. Daily counts for turtles, dolphins, boats and fish traps during the 2008 dugong surveys.

Power and sail boats were associated with tourism and, although infrequently

seen, were observed in very localized areas within the park or mainland near tourist

businesses.

Artisanal fishing boats, most under sail power, were scattered throughout the

bay. Many were supporting shore based net seining operations. The vessels were

often observed many hundreds of meters offshore of the mainland, shoals, or islands

with seine haulers back on shore with lines for each net end (5 or 6 people per end).

These hauls were reported to require about 5 hours of effort (L. Muaves/ WWF, pers.

comm. May 2008). It is obvious from the map distributions (Figures 4, A-2, and A-3)

that fishers are spread well beyond the shallow nearshore seining sites. Many of these

sightings could have been traveling boats as we did not record whether a boat was

actively fishing or travelling.

Shore based fish traps, or gamboas were historically set up for long term use and

were seen in a few specific spots but were not consistently reported by the aerial

observers. We noticed that none of them appeared to be tended and most did not have

nets attached to the structure. There are reports that they are seldom used any more in

this region (H. Motta / WWF, May 2008, pers. comm.).

Transects: With a 521 km flight path and 800 m search zone below the aircraft

(400 m per side), an estimated 416 km2 were searched by observers which represents

24% of the 1749 km2 total area.

DATE Turtle Dolphin Sail boat

Power boat

Artisanal Fish boat

Fish trap / Gamboas

May 25 39 NA NA NA NA May 26 36 NA NA NA NA May 27 54 13 May 28 44 8 May 29 48 5

19 

Data screening indicated that 400m was an appropriate distance for right-

truncation of the data; this eliminated about 5% of the sightings from the analysis.

Although there was evidence of heaping at angles that were multiples of 5, grouping the

data did not improve the apparent fit of detection functions, so exact distances were

used. Model selection resulted in a model with the hazard rate key function with no

series expansion term. Models which included covariates for animal type, observer,

survey day, and cluster size were not improvements over the simpler model with only

distance. Figure 9 shows a histogram of the observed detection distances with a plot of

the detection function superimposed.

0 100 200 300 400

0.0

0.2

0.4

0.6

0.8

1.0

Distance

Det

ectio

n pr

obab

ility

Figure 9. Histogram of detection frequencies as a function of distance for combined sightings of dugongs, sea turtles and dolphins, rescaled so that. A plot of the detection function fitted to the data is superimposed.

The expected cluster size was estimated two ways; a size-biased regression

method and as the mean cluster size for all groups seen within 400m of the adjusted

centerline. Although there was support for use of the regression to correct the expected

20 

group size for size-bias, the small sample size (17 clusters) reduced confidence in the

regression results and made decision difficult. Further, the percent of the overall

coefficient of variation of the density estimate due to cluster size estimation component

was over 50% for both methods. Thus, we present abundance estimates incorporating

both methods of estimation of cluster size. For the size-biased regression method,

)(ˆ sE = 1.85 (se=0.600) and the resulting abundance estimate for the study site was 112

dugongs, with a 95% confidence interval of (53,237). For the mean cluster size for all

groups seen within 400m of the adjusted centerline, )(ˆ sE = 7.65 (se=0.4.27) and the

resulting abundance estimate for the study site was 463 dugongs with a 95%

confidence interval of (155, 1378).

Abundance

Size Biased Cluster

Method

Density

(size biased)

(Area =1749km2)

Abundance

Mean Cluster Size

Density

(mean cluster)

(Area =1749km2)

112 0.06 463 0.264

DISCUSSION

Several dugong aerial surveys have been conducted over the Bazaruto Bay area

but have varied in one degree or another in terms of flight path, transect spacing,

frequency etc., (Cockcroft et al. 2008; Mackie 1999, 2001; Dutton 1998).

Generally the surveys performed by Cockcroft and Guissamolo in 2006 and

2007 were similar to the 2008 surveys reported here. Their area of coverage was

smaller but closer in extent to the 2008 work and included spacing between transects of

almost 4 km (2 nautical miles). The lengths of transects varied depending on the coastal

orientation and water depth, etc. as did ours. They flew 24 surveys over about 1.5

years with flight paths ranging from 222 km to 450 km. The total counts of dugongs

from their surveys showed the high variability we experienced and ranged from 3 to 69

dugongs, the maximum well below our maximum count of 135. Their more extensive

dataset resulted in an estimate of about 250 dugongs (global abundance of 247) which

is within the range of our two estimates of 112 and 463. Their densities using truncated

21 

distance data ranged from 0.004 to 0.09 dugongs/ km2 as compared to our two methods

for the three survey days yielding 0.06 and 0.26 dugongs / km2.

Given that aerial surveys previous to 2006 were of varying methods, spatial

scales, and timing, combined with the reality of the high variability in dugong sightings

between days in the BANP region, we cannot state that dugongs are increasing but that

there are numbers higher than previously reported in the literature. In addition to

current ecosystem stewardship policies, these recent observations should provide

significant motivation for Mozambique to continue implementation of protection for these

animals and their habitats. Our sighting of one large group of over 70 dugongs within

close proximity of a local shark fishing net (outside of the BANP) emphasized a current

fishery related threat to local dugongs.

Dutton (2004) described what he believed to be a decline in dugongs in

Mozambique. He reported that the southernmost population of Africa’s dugong at

Inhaca, once numbering about 20 is extinct, with small numbers being reported at

Inhambane. He also stated that the based on interviews and visits that the dugong was

extinct on Mozambique’s northern coast. He stated that an intensification of large mesh

gill-netting from 1976, coupled with lack of law enforcement, was the principal cause of

the decline of dugongs in Mozambique.

WWF has also expressed concerns of this species and associated habitats within

the region. They have particular concerns regarding a predicted fisheries collapse due

to overharvesting within in the region. Several corresponding problems can result from

this including a) destruction of seagrass beds underlying netting areas, reducing forage

available to dugongs and turtles; b) direct entanglement of dugongs by netting

operations; and c) anticipated collapse of fishery with reduced productivity (catch)

resulting in a revival of human harvest of dugongs and sea turtles for local subsistence.

Recommendations for future data collection efforts

The distance sampling method proved to work well in this application with a

reasonable amount of survey effort. However, there are a few key assumptions that

22 

should be tested in the future if resources become available. The first among these is

the assumption that all animals are seen on the survey line (g(0) = 1). This is a well-

known problem in distance sampling of marine mammals that occurs when animals that

are present are sometimes not available for observation during the period when the

observer passes a given point on the line (termed visibility bias, Pollock and Kendall

1987). Current investigations are exploring ways to correct distance sampling results

for this problem (e.g., Pollock et al. 2006). It would also be useful to explore the

assumption that combined sightings of dugongs, dolphins and sea turtles provides a

good approximation of the detectability of dugongs alone. An alternative would be to

increase the survey effort to the point where dugong sightings alone could be used to

model the detection function; 60-80 detections (of clusters, not individual dugongs) is

recommended as the minimum for reliable fitting of detection functions (Buckland et al.

2001). If future survey effort is available, repeated surveys should be conducted close

together in time to allow pooling of data for analysis.

Finally, because the estimation of expected cluster size has a large effect on the

abundance estimate obtained from distance sampling, and because of the large amount

of uncertainty due to this component, it may be optimal to put more effort into obtaining

a better estimation of cluster size. One method of doing this might be to routinely

collect cluster size data whenever dugong groups are observed by WWF scientists or

BANP rangers during boating transits in the bay. However, it is important to make sure

that size-bias is not creeping into the estimate (for example, if larger groups were more

likely to be encountered than smaller groups). Covariates such as distance of group

when first detected, sighting cues, season and geographic location should also be

recorded. These data would then be available to refine future dugong abundance

estimates based on distance sampling.

ACKNOWLEDGMENTS Alice Costa, WWF, Maputo provided all the logistical planning support for this project and Helena Motta took the helm in the project implementation in the field. Observers included Lara Mauves and Mario Fumo (WWF, Vilankulos),Thomas Chibate (Bazaruto Archipeligo National Park), Helena Motta (WWF-Mozambique) and Jane Provancha (Dynamac-NASA, USA). Recorders included Jose Rafael and Jane Provancha. David LePoidevin of Mission Aviation Friendship, Nampula, Mozambique provided excellent support and skills in daily piloting. NASA

23 

KSC Life Science Services contract supported the loan of manatee ArcPad software and custom data entry form for this dugong project. Resa Cancro provided guidance and support in data screening and output development. We thank Dr. Carlton Hall and Doug Scheidt for loan of the associated program and hardware. Special thanks to Helena Motta and the WWF staff at Maputo and Vilankulos for their hospitality, perseverance and patience. PERTINENT LITERATURE Buckland, S. T., D. R. Anderson, K. P. Burnham, J. L. Laake, D.L. Borchers, and L. Thomas. 2001. Introduction to distance sampling. Oxford University Press, Oxford, United Kingdom.

Burnham, K. P. and D. R. Anderson. 1998. Model selection and inference: a practical information-theoretic approach. Springer-Verlag, New York, New York, USA.

Cockcroft, V., A. Guissamulo, and K. Findlay. 2008. DUGONGS (DUGONG DUGON) OF THE BAZARUTO ARCHIPELAGO, MOZAMBIQUE, JANUARY 2008.

Cumming, D.H.M., Mackie, C.S., Dutton P., and Magane, S. 1995. Aerial census of dugongs, dolphins and turtles in the proposed Bazaruto National Park, Mozambique: April 1995. WWF Project Paper. MZ00006. 9pp.

Dutton, T.P., and R. Zolho. 1990. Conservation Master Plan for Sustainable Development of the Bazaruto Archipelago.. WWF report 1: 90 pp.

Dutton, T.P. 1993. Past and present status of dugong in the Bazaruto Archipelago and other known habitats on the Mozambique Coast.. Report for the DNFFB, 1: 3 pp.

Dutton, T.P. 1997. Mermaids in distress.. African Wildlife 51(6): 22-23.

Dutton, T.P. 1998. East African dugongs disappearing. Sirenews No. 29. 1p.

Dutton, P. 2004. Dugong Population Trends in the Bazaruto Archipelago National Park, Mozambique, 1990-2003. Sirenews 41, April 2004, pp. 12-14

Guissamelo, A., and V. Cockcroft. 1997. Aerial survey of dugong in the Bazaruto Archipelago. Report for the DNFFB.

Guissamulo, A. 2006. Dugong Survey. APRIL 2006. Environmental Impact Assessment for offshore exploration in blocks 16 & 19, Inhambane an Sofala Provinces, Mozambique. SASOL PETROLEUM SOFALA LIMITADA (SASOL) AND ENPERESA. NACIONAL DE HIDROCARBONETOS, E.P. (ENH). 27pp.

Jartley, H. _Nelson, _ Mayeux, and _Gregoire. 2008. The Assessment African Protected Areas: A characterization biodiversity ecosystem value. JRC Scientific and Technical Reports. www. European Commission.

Hughes, G.R. 1969. Dugong status survey in Mozambique. World Wildlife Yearbook (WWF) 1969: 137-139.

24 

Hughes, G.R. and Oxley-Oxland, R. (1971) A survey of dugong, Dugong dugon in and around Antonio Eres, Northern Mocambique. Biol. Conserv. 3:299-301.

Korrubel, J., and V. Cockcroft. 1998. Mermaids and Sirens. Divestyle 34: 36 pp.

Mackie, C. S. 2001. Aerial Census of Dugongs, Dolphins, and Turtles in the Bazaruto National Park, Mozambique: May, 2001. Eastern African Marine Ecoregion Programme WWF Project No: MZ0006, WWF SARPO, P O Box CY1409, Causeway Harare, Zimbabwe

Mackie, C.S. 1999. Aerial census of dugongs, dolphins and turtles in the proposed Bazaruto National Park, Mozambique: April 1999. WWF Project Paper. WWF Southern Africa Regional Programme Office. Harare.

Mackie, C.S., A. Correia, S. Magane, and L. Zivane. 1999. Aerial census of dugongs, dolphins and turtles in the proposed Greater Bazaruto National Park, Mozambique. WWF Project No. MZ0006. Report 1: 6 pp.

Mackie, C.S., A. Guissamelo, D. Nhantumbo, and C. Bento. 2001. Aerial census of dugongs, dolphins and turtles in the proposed Greater Bazaruto National Park, Mozambique.. WWF (Harare) Project No. MZ0006, Report 1: 12 pp.

Marques, F. F. C., and S. T Buckland. 2004. Covariate models for the detection function. In Advanced distance sampling.pp31-47. Buckland, S. T., D. R. Anderson, K. P. Burnham, J. L. Laake, D.L. Borchers, and L. Thomas (eds). Oxford University Press, Oxford, United Kingdom.

Marsh, H. H. Penrose, C. Eros, and J. Hugues 2006. Dugong Status Report and Action Plans for Countries and Territories Compiled by: HELENE MARSH, Action Plan Coordinator, IUCN/SSC Sirenia Specialist Group, HELEN Marshe- Compiler, UNEP/DEWA/RS.02-1 ISBN 92-807-2130-5

Management Plan Bazaruto Archipelago National Park 2002 – 2006. 2002. MAPUTO, MZ. Ministério do Turísmo DIRECÇÃO NACIONAL DAS ÁREAS DE CONSERVAÇÃO

Pollock, K. H. and W. L. Kendall. 1987. Visibility bias in aerial surveys: a review of estimation procedures. Journal of Wildlife Management 51: 502-510.

Pollock, K. H., H. D. Marsh, I. R. Lawler, and M. W. Alldredge. 2006. Estimating animal abundance in heterogeneous environments: An application to aerial surveys for Dugongs. Journal of Wildlife Management 70: 255-262.

Pusineri, C., Kiszka, J and Muir, C, 2008. The dugong in eastern Africa: balancing on the brink.. Sirenews No. 49, April, 2008.

Thomas, L., Laake, J.L., Strindberg, S., Marques, F.F.C., Buckland, S.T., Borchers, D.L., Anderson, D.R., Burnham, K.P., Hedley, S.L., Pollard, J.H., Bishop, J.R.B. and Marques, T.A. 2006. Distance 5.0. Release 2. Research Unit for Wildlife Population Assessment, University of St. Andrews, UK. http://www.ruwpa.st-and.ac.uk/distance.

25 

Appendix – Additional Distribution Maps

26 

Figure A-1. Daily distribution of marine turtles sighted during the official survey dates of 27- 29 May 2008.

Figure A-2. Daily distribution of boats (artisanal, touristic) and fish traps during the official survey dates of 27- 29 May 2008

27 

Figure A-3. Daily distributions of only artisanal fishing boats during official aerial surveys conducted 27-29 May

28 

Figure A-4. Daily dolphin distributions during official aerial surveys conducted 27-29 May 2008.

29 

Figure A‐5.  Example of summary sheet for each aerial survey. 

BNMP Dugong Surveys

DATE _________________________________Flight No __________________ Aircraft Type __________________________Aircraft Tail No _____________ GPS Test check____________________________________ Observers___________________________________ Pilot _____________________ Weather _______________________________________________________________ Wind _________________________________ Water Surface (seastate): ________________________________________________ Turbidity notes:_________________________________________________________ Start Survey Time _________________ Stop Survey Time _________________ Total Survey Time_____________________ Total Flight Time___________________ Condtions Summary (Excellent, good, fair, poor):___________________________ ______________________________________________________________________ Average Altitude ______________________ Average Speed _________________________ Flight Route: _________________________ Flight Summary:

Total Dugongs Total Turtles

NOTES:

________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

______________________________________________ Data entry date________________ Initials______________ Revision May28 2008